Neural network-based speech recognition technology for voice assistants and IoT devices
WhisPro™ is an AVS qualified neural network based keyword spotting (KWS) software package, allowing...
Audio Front-End voice processing software package providing enhanced speech clarity for voice-enabled devices
Clear speech is fundamental to communications, whether between humans or human-to-machine as part of a Voice User Interface. Virtual communication has become ever more critical in our hybrid, distributed world, whether over a conferencing system, headset, or even a smart speaker. Meanwhile, voice has become the primary user interface for services such as cloud-based conversational assistants and voice activated IoT devices. These all require a clean voice signal in any given environment to ensure accuracy and intelligibility by smart speakers, Bluetooth connectivity speakers, conferencing devices, headsets, smart appliances, IIoT devices, automotive, and smart home systems.
ClearVox incorporates advanced algorithms that cope with different acoustic scenarios and microphone configurations, including optimized software for speaker direction of arrival (DOA), multi-mic beamforming, noise reduction, and acoustic echo cancellation (AEC).
ClearVox AFE has proven quality, having passed relevant tests in a third-party lab for Microsoft Teams certification for audio-video conferencing systems. For those building a Voice UI-enabled device, ClearVox AFE is an AVS qualified Audio Front End software solution, ready for integration with Alexa-enabled devices and optimized for Smart Speakers, Soundbars, and Smart Home IoT devices.
Benefits
Software suite of advanced voice input algorithms for voice communication devices. Provides confidence and quality for device makers
Configurable and modular, lets silicon vendors and OEMs customize for product needs like number of mics, mics geometry and echo tail length
Provides more than 15dB SNR improvement in most conditions, DOA typically accurate to 15° (max 5°), far field voice pick-up, and enables barge-in during music playback for voice UIs
Unique software architecture minimizes computational workload for cost and power efficiency
Main Features
- Incorporates multiple algorithms such as software VAD (Voice Activity Detection), direction of arrival, beamforming, noise reduction and acoustic echo cancellation
- Configurable for any voice-enabled application, from Smart Home appliances up to conferencing and automotive
- Accompanying SDK is available for customers to configure and adapt ClearVox AFE to their product specifications, including number and geometry of microphones, echo tail length, and more
- Ceva provides remote and on-site support for Microsoft Teams certification
- ClearVox AFE operates in tandem with WhisPro, or any speech recognition engine, offering a complete CEVA DSP-based software solution for speech recognition
- Providing state of the art noise reduction and AEC for barge in capability
- Ensured to work together with minimal computational resources and power consumption with high recognition rate in noisy environments in near-field and far-field scenarios
- ClearVox AFE is available today for Ceva’s DSPs: Ceva-BX1, Ceva-BX2 and SensPro2 family, as well as Arm CPUs
Block Diagram
Ceva Voice-Control Enabled Solutions for IoT
End-to-end voice control solutions for wearable, mobile, smart home, and automotive markets
Based on Ceva’s Sound Software packages: ClearVox AFE noise reduction and WhisPro speech recognition Artificial Intelligence.
Available for licensing for Ceva-BX1, Ceva-BX2 and SensPro family, as well as for ARM CPUs.
